MECHANISM OF
May, 1949
[CONTRIBUTION FROM
THE
THE
LITHIUMALUMINUM HYDRIDEREACTIONS
1675
GEORGE H E R ~ E RJONES T LABORATORY, THEUNIVERSITY OF CHICAGO]
Mechanism of Lithium Aluminum Hydride Reactions BY LLOYDmr,TREVOY' AND WELDON G. BROWN With few exceptions the known reactions of basis of infrared absorption spectra,2 relative lithium aluminum hydride with organic com- rates of dehydration3 and of h y d r ~ g e n a t i o n , ~ pounds consist essentially of the displacement of a appears to be reasonably certain. The epoxides strongly electronegative element (oxygen, nitro- are presumed to have the cis configuration and gen or halogen) on carbon by hydrogen. Non- inversion will have been demonstrated if the inpolar groups, e.g., isolated double bonds, are not coming hydrogen atoms take up positions trans affected. It is therefore reasonable to suppose to the hydroxyl groups in the reduction products. The isomeric alcohols are characterized by dif that the reactions proceed by polar mechanisms. Moreover, the hydride reactions thus far reported ferences in physical constants which, with the have certain features in common with the more exception of the significantly lower boiling points familiar nucleophilic displacement reactions on of the trans isomers, are relatively minor. Identicarbon and a common basis in mechanism is in- fication of the reduction products on the basis of these differences (cf. Table I) is necessarily dicated. The simplest assumption would be that hydride tenuous in view of the fact that neither the reducion, furnished by dissociation of the complex tion products nor reference samples of the indihydride, functions as a nucleophilic reagent. vidual isomers could be obtained in pure form But apart from the lack of evidence for such because of decomposition during distillation. dissociation, a mechanism based upon the as- The weight of evidence is in favor of the trans sumption of free hydride ions would not easily configuration for the reduction product in each account for the characteristic differences in the instance; the observed boiling points and densireducing properties of different complex hydrides, ties support such an assignment while the refrace.g., aluminohydrides vs. borohydrides. It ap- tive indices and melting points are indecisive. The tentative identification of the reduction pears more likely that the reactant is actually a series of complex aluminohydride ions, A l X n - product from 1,2-dimethyl-l,2-epoxycyclopentane H4- %-, where n progresses from zero to four during as trans-l,2-dimethylcyclopentanolis substantithe course of a reaction, which act as carriers for ated by the observation that its melting point is hydride ion. The first phase of the reaction with strongly depressed by admixture of the cis referan epoxide, for example, might then be formulated ence sample but not by the trans reference sample. Similar confirmatory evidence for the dimethylas cyclohexanols, which presented the greatest difv v c-oficulty in purification, is lacking. AlHd-
+
c\ I
C/
A
0
+ AlHs
+ HC-I I
If the reactions proceed via the usual bimolecular displacement mechanism, as the above formulation is intended to suggest, there are two experimental tests which may be conveniently applied. One is the test as to whether inversion of configuration occurs on the carbon atom attacked. The other is the correlation with respect to relative reactivities and with respect to the point of attack, if two alternate paths are possible, with other reactions known to proceed by such a mechanism. An approach based on reaction kinetics does not appear to be feasible. Suitable test cases for the experimental demonstration of inversion are provided by 1,2-dimethyl-l,2-epoxycyclopentaneand 1,2-dimethyl1,2-epoxycyclohexane which we i k d to be reduced by lithium aluminum hydride to the corresponding dimethylcyclanols. These tertiary alcohols are each known in two stereoisomeric forms, and the assignment of configuration on the (1) Standard Oil Company (Indiana) Research Fellow, 1947. Present address: University of Saskatchewan, Saskatoon, Saskatchewan.
TABLE I PHYSICAL CONSTANTS
O F ISOMERIC DIMETHYLCYCLANOLS cis Form trans Form This This Obs. red. Lit. work Lit. work product 1,2-Dimethylcyclopentanol
B. p.
69.6'/25'
M.p. 28.9' 7lD
dgQi
B. p.
1.4523b 0.9088' 82.8'/25'
M.p. 23.2' 1.464Sd 0.9250'
62'/17 58.4'/25" 24-26 27.3a 1.4516 1.4463b 0.9005"
51°/17 24-25 1.4506 0.9094
53'/17 23-24' 1.4493 0.9094
1,2-Dirnethylcyclohexanol 74 .0°/25' 62.8'/17 35'/2.5 13.2' -3.5 to - 1
1.4613d 0.9187'
1.4587 0.9189
35'/2.5 -8 t o -'7 1.4594 0.9174
a Chiurdoglu, ref. 3 : the literature density values given here are for d304. Van Rysselberge, Bull. SOC. chim. Belg., 35, 316(1926). The refractive indices given by this author are for nZ5D; all other values appearing in this table are n%. Chiurdoglu, ibid., 47, 245 (1938). Chiurdoglu, ibid., 50, ZO(1941). e Mi%ed rn. p. with cis-1,2-dimethylcyclopentanol,-8 t o -6 ; mixed m. p. with trans form, 22-24'. Mixed m. p. with trans-l,Z-dimethylcyclohexanol, -6 to -5'.
*
(2) Chiurdoglu. Barchewitz and Freymann, Bull. 47, 453 (1938). (3) Chiurdoglu, ibid., 44, 631 (1933).
SOC.
ckim. Bclg.,
1676
LLOYD W. TREVOY AND WELDON G. BROWN
Vol. 71
With respect to directive effects in the bimolecular displacement reactions of unsymmetrical epoxides, the reactions of epoxides with sodium ethyl malonate are unquestionably of this type4 and will serve as a basis of comparison. Styrene oxide is attacked both by lithium aluminum hydride6 and by sodium ethyl malonate6 preferentially a t the terminal carbon atom. That this is also true for 3,4-epoxy-l-butene and for epichlorohydrin is shown by the predominant formation of secondary alcohols in the hydride reductions (cf. Table 11) in conjunction with the identification of the malonic ester condensation products in
The replacement of chlorine by hydrogen, which accompanies (at a slower rate) the reduction of the epoxide ring of epichlorohydrin, is a normal type of reaction for lithium aluminum hydride. The examples previously reported, together with the further results given in Table 111,show clearly that the trends in the reactivity of mono-halides exhibit the familiar pattern of nucleophilic displacements. A detailed discussion in substantiation of this conclusion does not appear to be necessary. However, the apparently anomalous behavior of diarylmethyl halides and of dihalides requires comment. The abnormality shown by both diphenylbroTABLE I1 momethane and 9-bromofluorene is the formation REDUCTION O F EPOXIDES BY LITHIUM ALUMINUMHYDRIDE^ of substantial proportions of the corresponding Product Yield, % tetraarylethanes. An explanation is to be found Styrene oxide a-Phenylethanol 75 in the observation that fluorene reacts with lithium 3,4-Epoxy-l-butene 1-Butene-3-01 58 aluminum hydride to form an organometallic 1-Butene-4-01 13 compound, demonstrated through the isolation Benzalacetophenone1,3-Diphenylpropaneof fluorene-9-carboxylic acid after carbonation. oxide 1,2-diol 79 Presumably diphenylmethane is also sufficiently Epichlorohydrin Propanol-2 88 acidic to react similarly although this was not inCyclohexeneoxide Cyclohexanol 91 vestigated. It is clear that the metallic derivatives, reacting with more of the halide, could give a Reductions carried out in diethyl ether a t its boiling point, except epichlorohydrinowhich was reduced in di- rise to the observed coupling products. ethylcarbitol solution a t 25 , Reaction times varied With 1,2-dihalides the normal displacement refrom twenty t o forty-five minutes. action is wholly or partly suppressed in favor of the former case by Russell and VanderWerfIs olefin formation. The latter is the principal reand in the latter case by Traube and Lehmann.' action for 1,2-diiodides and for those 1,2-dibroThe isolation of some primary alcohol, repre- mides which can give rise to a conjugated olefin senting attack by hydride a t the secondary car- (e.g., butadiene, styrene, stilbene) by the eliminabon atom of 3,4-expoxy-l-butene, does not in- tion of bromine. It is of interest to note that validate the comparison since the observed yield lI4-dibromo-2-butene, which could furnish butain the malonic ester reaction (64%) does not pre- diene by bromine elimination, reacts instead by a clude a minor reaction by the alternate path. normal displacement which is greatly facilitated In fact our findings are entirely consistent with by the allyl-type activation. Olefin formation the results of Bartlett and Rosss for the alkaline from 1,2-dihalides can be regarded as nucleophilic methanolysis of 3,4-epoxy-l-butene and these displacements on halogen, as Winstein, Pressman authors present evidence to show that the reac- and Youngg have shown for the halogen eliminations a t the primary and secondary positions are tion brought about by iodide ion. The inversion of configuration, which should both of the bimolecular displacement type. The explanation for concurrent attack a t the secondary accompany the normal displacement of halogen position (allyl-type activation) would apply with by hydrogen, could be observed experimentally equal force to styrene oxide for which only one only in the case of a tertiary halide. However, reduction product has been reported. I n a re- the tendency to undergo normal displacement is examination of this point we obtained a small lowest for this type, and under forcing conditions, fraction of material boiling higher than a-phenyl- they react chiefly by dehydrohalogenation. I n ethanol, in the correct range for p-phenylethanol, principle a t least, the demonstration of inversion but too small for further fractionation or identifi- could be carried through for the case of a secondcation. A substituent on the p-carbon atom of ary halide reacting with lithium aluminum styrene oxide evidently shifts the main attack deuteride. In pursuance of the stereochemical aspects of to the a-position, as is shown by the formation of 1,3-diphenylpropane-l12-diolin high yield hydride reductions, some observations on the from benzalacetophenone oxide. In this reaction isomeric composition of the products from the the reduction of the carbonyl group doubtless reduction of 1,2-diketones (cf. Table IV) may be noted. Directive effects appear, and are enprecedes opening of the epoxide ring. hanced somewhat by working a t a low tempera(4) Grigsby, Hind, Chanley and Westheimer, THISJOURNAL, 64, ture, but in general the products appear to be 2606 (1942). (5) Nystrom and Brown, ibid., 70, 3738 (1948). mixtures of the various possible stereoisomers. (6) Russell and VanderWerf, i b i d . , 69, 11 (1947). Possibly the most interesting feature of these (7) Traube and Lehmann, Bcr., S4, 1977 (1901). (8) Bartlett and Ross, THISJOURNAL, 70, 926 (1948).
(9) Winstein, Pressman and Young, i b i d . , 61, 1645 (1939).
May, 1949
MECHANISM OF
THE
LITHIUM ALUMINUM HYDRIDEREACTIONS
1677
cal constants the reduction product is believed to be the trans isomer. Reduction of Other Epoxides.-Styrene oxide was furnished through the courtesy of the Dow Chemical Company; 3,4-epoxy-l-butene through the courtesy of the Columbia Chemicals Division, Pittsburgh Plate Glass Company. The reduction of these, and the other epoxides shown in Table 11, has been carried out by procedures similar to those outlined above. All of these reductions were rapid at the temperature of boiling ether. A reduction of 3,4-epoxy-l-butene, carried out at -40", also occurred rapidly and the product had substantially the same composition as that obtained at the higher temperature. All of the products shown in Table I1 are known compounds and the identity was confirmed wherever possible by the preparation of known derivatives. Confirmation was proof the identity of 1,3-diphenylpropane-l,2diol1~ vided by periodic acid oxidation which formed benzaldehyde and phenylacetaldehyde, identified as their dinitrophenylhydrazones. Reduction of Halides.-Experiments relating t o the reduction of halogen compounds are summarized in Table 111. As noted, various higher-boiling solvents were employed to effect reduction of the more resistant types a t elevated temperatures. These experiments were carried out with quantities of halide in the range 0.1 t o 0.4 mole Reduction of 1,2-Dimethyl-l,2-epoxycyclopentane.and with quantities of lithium aluminum hydride 30 to The epoxide was prepared by the procedure outlined by 100% in excess of theoretical. Hydrocarbon products Bartlett and Bavleyll except that the intermediate, 1,2- were identiiied on the basis of physical constants except dimethylcyclopentanol-1 , was subjected to fractional dis- trans-butene-2, which was converted to the dibromide, tillation through a Podbielniak Heligrid column in order meso-2,3-dibromobutane, b. p. 62-65' (32 mm.), n Z o D to secure reference sample of the cis and trans isomers. 1.5110; lit." b. p. 66.2-66.6' (30 mm.), n Z o D 1.5116. Middle fractions from this distillation were then used in Hydrocinnamyl $lcohol was identified s; the p-nitrobenthe further steps of the synthesis. Our product had a zoate, m. p. 46 ; lit.16 m. p. 47-48 The reduction boiling point, 123.5-124' (741 mm.) ; the boiling poini product from 9-bromophenacyl bromide, a-(p-bromoreported by the above-mentioned authors, 120-122.5 phenyl) -eihanol, was converted to the N-phenylcarbamate, (20 mm.), is evidently misprinted in the original article. m. p. 103 ; lit.1Pm. p. 103-104". The reduction of 20 g. (0.178 mole) of the epoxide by The reduction of 1,2,3,4-tetrabromobutane at 35" 0.118 mole of lithium aluminum hydride in diethyl ether furnished an unsaturated liquid product of acrid odor, consolution took place relatively slowly. Spontaneous re- taining bromine, and boiling over the range 90-118' fluxing continued for two hours and thereafter the mixture (748 mm.). At 65', in tetrahydrofuran, the reaction of was heated under reflux for an additional hour. The ex- the tetrabromide with lithium aluminum hydride produced cess hydride was destroyed by the addition of water, alkali butadiene. Again in the case of 2,3diiodopropanol-1 a was added, and the product was worked up in the usual pure product was not isolated; the limited scale of the exway. On distillation a t 17 mm. pressure using a small periment (0.05 mole) did not permit a clean separation of wire-spiral column, the material was separated into un- allyl alcohol, indicated by bromine titration to be the chief changed epoxide, intermediate fractions, a main fraction product, from other substances of comparable volatility of 8.2 g., b.p. 53-54' (17 mm.), and an insignificant which were present. residue of higher boiling material. On redistillation of the In the reduction of 9-bromofluorene, an orange color main fraction a forerun of lower boiling material was appeared on the f i s t addition of the halide, disappeared separated ; the remainder of the distillate was character- as the reaction progressed, and reappeared some thirty ized by the data given in Table I and is thus identified as minutes after completing the addition. Dibiphenylenetrans-~,2-dimethylcyclopentanol-l. ethane, m. p. 247-247.5' after recrystallization from benReduction of 1,2-Dimethyl-1,2-epoxycyclohexane,- zene-ethanol (lit." m. p. 247'), separated from a concenThe compound (b. p. 151.5-153.5' (747 mm.), T Z ~ ~ Dtrated ether extract of the product. The residue from the 1.4501) was prepared in 407, yield by the perbenzoic mother liquor, taken up in petroleum ether and subjected acid oxidation of 1,2-dimethylcyclohexene-1 following the to chromatographic adsorption, was separated into fluoprocedure of Nametkin and Delektorsky.lz Reduction rene, a further small quantity of dibiphenyleneethane, and occurred readily on the dropwise addition of 11 g. of the a small amount of an orange-colored substance which was compound, diluted with ether, to 100 ml. of a solution con- not identiiied. taining 0.118 mole of lithium aluminum hydride, after To confirm the intermediate formation of a metallic dewhich the mixture was allowed to stand for twenty rivative of fluorene, suggested by the above-mentioned minutes. Difficulty was experienced in purifying the color phenomena, a mixture of fluorene (0.15 mole) and product because of the tendency to dehydrate during pro- lithium aluminum hydride (0.0376 mole) in tetrahydrolonged distillations. The material from the first distilla; furan (nitrogen atmosphere) was heated a t 65" for fortytion using a short wire-spiral column, 8.2 g., b. p. 40-41 two hours at which time the evolution of gas had slackened (2.5 mm.), was redistilled from a Claisen flask and the and a deep red color had developed. Carbon dioxide was resulting product was characterized by the physical con- then introduced and the color gradually changed to orange. stants shown in Table I. The melting points were not The greater part of the fluorene was recovered but fluorenesharp and the product was probably contaminated with 9-carboxylic acid, m. p. 225-226" (lit.'* m. p. 226 or 232"), olefin. A specimen of trans-l,2-dimethylcyclohexanol-l, was isolated in the amount of 4.6 g. (0.022 mole). separated from a synthetic mixture of the cis and trans (13) Levy and Dvoleitzka-Gombinska, Bull. SOC. chim., [4] 49, forms by means of the Podbielniak column, behaved (1931). similarly, and on the basis of the close agreement in physi- 1772 (14) Kharasch, Lambert and Urry, J . Ow. Chcm., 10, 303 (1945).
results is the close relationship, in composition of the products, with catalytic hydrogenation and the sharp contrast in certain instances (e.g., camphorquinone, camphor) with the products of reduction by active metals. This points toward a common steric element in the attack by a complex hydride ion on the one hand and the attack by a catalyst-hydrogen complex on the other. I n comparison with lithium aluminum hydride, the borohydrides of lithium and sodium differ principally in their lesser aggressiveness. At ordinary temperatures, the action of the borohydrides is confined to the more readily reducible types of groups.1o It is highly probable that the mechanism of their action is essentially similar, and that their lower reactivity is a consequence of the greater stability of the borohydride ion and of the reluctance with which it transmits a hydride fragment to an electron-deficient center. Experimental
.
(10) Chaikin and Brown, THISJOURNAL, TO, 122 (1948); Chaikin, Ph.D. Dissertation, The University of Chicago, 1948. (11) Bartlett 8nd Bavley, THISJOURNAL, SO, 2416 (1938). (12 Nametkin and Delektorsky, Bcr., 67,585 (1924).
48, 1111 (1926). (15) Kirner, THISJOURNAL, (16) Quelet, Bull. SOC. chim., [4] 46, 88 (1929). 68, 379 (1946). (17) Pinck and Hilbert, THISJOURNAL, (18) Jeanes and Adams, ibid., 69, 2620 (1937).
LLOYD W. TREVOY AND WELDON G. BROWN
1678
Vol. 71
thirty minutes at 35'. The over-all yields were slightly TABLE I11 improved at the lower temperature ( c f . Table IV) but the REDUCTION OF HALIDES BY LITHIUM ALUMINUM HYDRIDE most marked improvement was in the optical purity of Time, Yield, optically active products. In no case, however, could the % hr. Product reduction be directed t o the formation of a single stereoI n Diethyl Ether, a t 35" isomer. The product from d-camphor at -80", representing the nearest approach, consisted chiefly of d ( Benzyl iodide 0.3. Toluene 86 isoborneol, m. p. 213", 3,5-dinitrobenzoate m. p. 139 , a-(p-Bromophen yl) p-Bromophenac yl [aIasD -30.6" (lit.'@m. p. 214", 3,5-dinitrobenzoatem. p. bromide 1 ethanol 85 139O, [ ~ ] * O D in methanol -32.30') and, as indicated by the low rotation, doubtless contained d ( +)-borneol. TriphenylchloroTriphenylmethane 98 The product from the reduction of d-2,3-camphorquinone, methane 0.3 when purified simply by recrystallization, furnished physical constants in reasonably good agreement with In Tetrahydrofuran, a t 35" literaturez0values for cis-2,3-camphaneglycol. But when Benzyl bromide 2 Toluene 78 purified by conversion to the isopropylidenz derivative 9-Bromofluorene 0 . 5 Fluorene 30 and regeneration, our product, m. p. 257 , [ a ] * Oin~ ethanol, -19.02' (lit.*Om. p. 254-256", [a]*'D in ethanol, Dibiphenyleneethane 34 -13.4'), had increased in specific rotatory power. Our In Tetrahydrofuran, a t 65' product was therefore not optically homogeneous. Cyclohexane-1,2-dione reacted with lithium aluminum 72 Benzyl chloride 0 . 5 Toluene hydride with the evolution of one mole of hydrogen gas n-Decane 72 1-Bromodecane 10 and formed, as the only identifiable product, cyclohexanol30 2-BromoBctanea 10 n -0cta n e a-one, m. p. 139-142' (lit.21 m. p. 138'), characterized as the phenylosazone, m. p. 152-153" (lit.**m. p. 153'). Styrene 49 w-Bromostyrene 19 24 No reaction fi-Fluorotoluene TABLE IV 40 Chlorobenzene 1-Chloro-2-iodoREDUCTION OF KETONES benzene' 0.5 Yield at Ethyl @-chloroethyl Ethyl a,@-dichloroProduct 35. -800 53 ethyl ether ether 0.5 Hydrobenzoin (meso) 81 90 Benzil Pentaerythrityl No reaction Isohydrobenzoin 5 bromide d( -)-Isoborneol 97' 97' d-Camphor Hydrocinnamyl Ethyl 2,3-dibromo-3d-2,3-Camphaneglycol 97' 98d d-2,3-Camphoralcohol 59 phenyl propionate 1 quinone 5 Trimethyleneglycol Diethyl bromocis-Acenaphthyleneglycol 15 13 Acenaphthenemalonate 22 quinone trans-AcenaphthyleneDiphenylmethane 38 Diphenylbromoglycol 45 50 25 methane 0.5 Tetraphenylethane Cyclohexane-l,271 0 . 5 Styrene Styrene dibromide Cyclohcxanol-2-one 41 dione meso-1,a-Diphenyla [(Y]:)D (in methanol, c IO) -24.87;. [CY]*'D 98 1,2-dibromoethane 4 . 5 trans-Stilbene -30.6 [ a ] 2 8 (in ~ ethanol, c 7) -14.6 * [aIz4D 17 -17.3'. Octene-1 1,2-Dibromo6ctanee 2 26 2-BromoGctane
-\-
-
..
.
In Diethylcarbitol, at 65 " trans-l,4-Dibromo-2trans-2-Butene butene 0.3
72
In Dibutylcarbitol, a t 65" 2-Bromo-2-methyl2-Methyloctenes octane 7
76
t-Butyl iodide a
In Dibutyl Ether, at 80' 3 Isobutane
29% recovered.
Isobutene 38% recovered. 22% recovered.
Reduction of 1,2-Diketones.-In the expectation that the stereochemical specificity of the hydride attack would be enhanced at a lower temperature, the reductions we;e carried out a t -80' (Dry Ice-bath) as well as a t 35 The solvent employed throughout was diethyl ether. At -80" the reactions were noticeably slower, requiring one to two hours for completion as compared with twenty to
.
. '
Summary A mechanism for reductions by lithium aluminum hydride, which involves a nucleophilic attack on carbon by complex hydride ions, is supported by (a) the occurrence of inversion in the reactions with bicyclic epoxides, (b) a comparison with sodium ethyl malonate with respect to the site of reaction in unsymmetrical epoxides, and (c) the behavior of organic halides toward the reagent. The reduction of 1,2-diketones by the hydride is shown to form products resembling, in isomeric composition, those formed in catalytic hydrogenation. CHICAGO, ILLINOIS
RECEIVED OCTOBER11, 1948
(19) Lipp, Bcr., 74, 8 (1941). (20) Rupe and Thommen, Helu. C k i m . Acta, 80, 942 (1947). (21) Halasz, A n n . ckim., 14, 362 (1940). (22) Bergmann and Gierth, A n n . , 448, 61 (1926).
